Sensors and Actuators A 301 (2020) 111693

Contents lists available at ScienceDirect

Sensors and Actuators A: Physical

journal homepage: www.elsevier.com/locate/sna

Structural monitoring system for proactive detection of corrosion and

coating failure
Jawwad Latif a , Zulfiqar A. Khan a,∗ , Keith Stokes b
a
Bournemouth University, Department of Design and Engineering, NanoCorr, Energy and Modelling (NCEM) Research Group, United Kingdom
b
University of Southampton, National Centre for Advanced Tribology at Southampton (nCATS), United Kingdom

a r t i c l e i n f o a b s t r a c t

Article history: Performance and availability of high priority structures, systems and products can be greatly affected
Received 3 July 2019 by corrosion damage. The application of protective coatings, frequent inspections and scheduled based
Received in revised form maintenance activities result in huge direct and indirect financial loss to organisations. The expeditious
17 September 2019
detection of coating failure and corrosion damage can result in precise and cost-effective condition-
Accepted 22 October 2019
based maintenance. Coating failure and corrosion phenomena are driven by complex multi-disciplinary
Available online 15 November 2019
parameters according to extensive research findings in the literature. State-of-the-art prognostic mod-
els proposed in recent years incorporate complex multi-disciplinary parameters, therefore a real-time
Keywords:
Corrosion sensors
prognostic monitoring system must acquire these complex parameters to allow accurate prediction. The
Electrochemistry work reported here covers the development of a real-time monitoring system by using micro-sensors
Prediction and includes the validation of the system through accelerated corrosion and coating failure testing. The
Prognostics system contains a remote terminal unit that includes a linear polarisation method for corrosion detection
Coating under the coating and a micro-strain gauge method for monitoring stress behaviour over the coating. The
software at a base station includes a graphical user interface and database to store parameters for further
processing and failure prediction. Real-time monitoring system can be applied to remote, stationary and
mobile assets to monitor mechanical and chemical changes within coating-substrate systems.
© 2019 Elsevier B.V. All rights reserved.

1. Introduction facial toughness, diffusion-induced stresses and residual stresses

are considered to be responsible for micro-cracks and adhesion
Critical structures, systems and products including aircraft, loss in protective coatings applied to metal structures [2–12]. Var-
automobiles, highway bridges, energy production, distribution ious maintenance activities are implemented to protect and repair
systems, and household appliances can be highly susceptible to the metal structures e.g. brushing, washing and recoating. The
corrosion damage due to excessive operational activities in uncon- maintenance of critical metal structures, therefore, can be costly
trolled environmental conditions. According to recent studies, the and time-consuming due to carrying out excessive inspection and
direct cost of corrosion damage is around 3 % of the national Gross recoating maintenance, coupled with any delay in productivity. The
Domestic Product [1]. Indirect costs, including the cost of operator most commonly practiced scheme for maintenance is referred to
activities, the overhead cost of delays, loss in productivity and any as Scheduled Based Maintenance (SBM) which is carried out at reg-
litigation issues are estimated as being equally costly. ular intervals independent of the condition of the structure. This
Durable organic coatings are applied to structures to provide a often results in unnecessary maintenance and has led to studies into
protective barrier from surrounding environmental conditions. The alternative, more cost effective, approaches over the past decade.
organic coatings are susceptible to numerous types of failure modes These studies have resulted in the implementation of Condition
including micro-cracking within the coating surface which allows Based Maintenance (CBM), where the condition of the structure
the corrosive species to diffuse into the coating-substrate interface indicates that maintenance is necessary. The approach to identify
to initiate corrosion damage [2]. According to recent experimental the need to implement such maintenance strategies ideally needs
investigations, parameters including interfacial roughness, inter- to be low cost, simple to install and operate. Sensor technology can
provide an efficient solution for condition monitoring of structures
operating at remote locations.
∗ Corresponding author.
The corrosion mechanism lies under electrochemistry con-
E-mail address: zkhan@bournemouth.ac.uk (Z.A. Khan).
cepts while residual stresses are mechanics parameters. Various

https://doi.org/10.1016/j.sna.2019.111693
0924-4247/© 2019 Elsevier B.V. All rights reserved.
2 J. Latif, Z.A. Khan and K. Stokes / Sensors and Actuators A 301 (2020) 111693

corrosion sensors have been developed and are characterised as which diffuse through the coating due to contraction and they
direct and indirect corrosion monitoring techniques. The direct can also generate blisters if they exceed a critical level [6]. Sev-
monitoring technique involves the measurement of potential or eral methods are applied to measure residual stresses which are
current resulting from corrosion/electrochemical reactions. Corro- characterised as Non-destructive, Semi destructive and Destructive
sion coupons, electrochemical impedance spectroscopy, electrical techniques [16]. X-ray diffraction, Neutron diffraction, Ultrasonic
resistance and potential measurement using linear polarisation and Barkhausen noise methods lie within the Non-destructive
are examples of direct corrosion monitoring techniques. These category. Hole-drilling, Ring core and Deep-hole methods are semi-
techniques are also termed as intrusive as they require access to destructive techniques while Sectioning and Contour methods are
the metal structure for corrosion measurement. Indirect corrosion destructive techniques used for residual stress measurement.
monitoring techniques are based on an outcome resulting from cor- The behaviour of tensile and compressive stresses in coatings
rosion reaction. Radiography and ultrasonic testing are common needs to be monitored in real time for efficient prediction of
indirect and non-intrusive corrosion monitoring techniques. Both potential structural damage. Although strain gauge sensors have
techniques have advantages and disadvantages. The corrosion rate been widely used in aerospace applications, bridge cables, mea-
from corrosion coupons is estimated by measuring the weight-loss surement of torque generated by motors, turbines, propellers or
over a particular length of time. It provides a reliable physical indi- wheels, the performance of strain gauges in monitoring corro-
cation of the overall uniform corrosion rate but is time-consuming sion and coating damage assessment needs to be investigated in
and cannot be used in real-time monitoring systems [13]. The elec- more detail. The residual stresses generated during physical vapour
trical resistance method measures the change in resistance caused deposition coating have been scaled to real-time using the strain
by metal loss. The rate of corrosion can be measured at any time and gauge sensor. The experimental data generated from strain gauge
as frequently as needed but requires calibration based on the prop- monitoring has been validated with predicted values of residual
erties of the structural material [14]. Electrochemical Impedance stress. The performance of strain gauges in monitoring the devel-
Spectroscopy is used to measure the corrosion rate of the metal opment of residual stresses in the coating-substrate system due
structure as well as any corrosion occurring under the coating. It to change in temperature has also been investigated [17]. In the
is a powerful method of determining corrosion rate, but it is very current work, strain gauge sensors have been used to acquire real-
sophisticated and prone to noise in the field environment. Lin- time measurements of residual stresses developing in the coating.
ear Polarisation Resistance (LPR) method is also a powerful tool The proposed structural monitoring system is based on a sensor
that provides a direct and instantaneous determination of corro- suite to detect electrochemical reactions using ␮LPR sensors and
sion rate which can be utilised for real-time structural monitoring. mechanical residual stresses using strain gauge sensors. In addi-
The disadvantage of using LPR technique, however, is that it is only tion, temperature and humidity sensors are included within the
effective in clean, aqueous electrolytes. suite.
The performance of ␮LPR sensors has been evaluated using an
accelerated corrosion test to ASTM G5-A [13]. From these tests it
was established that ␮LPR sensors can precisely measure the low
corrosion rate in corrosive atmospheric conditions. Using these
sensors, the NanoCorr Energy Modelling and Research (NCEM) 2. Proposed structural monitoring system
group have investigated corrosion damage within large vehicles at
The Tank Museum, Bovington, Dorset UK. The ␮LPR sensors were The Tank Museum, Bovington is located near the English Chan-
embedded on vehicles operating in both controlled and uncon- nel in Dorset where large military vehicles are exhibited and
trolled environment at remote locations [4,15]. The Tank Museum operated for various purposes. The presence of high amounts of
is located near the English Channel, hence a high amount of salin- salt particles in the atmosphere combined with wind speed, rain,
ity is present in the atmosphere. The average wind speed always relative humidity and temperature are causing a potential threat
remains greater than the threshold level for entrainment of salt to their structural integrity. It has been established in the litera-
particles. The relative humidity level also remains greater than the ture that the stresses within a coating control the compression and
critical threshold level throughout the calendar year. The valuable opening of micro-cracks that allows corrosive salt particles to dif-
vehicles operating in these harsh atmospheric conditions are highly fuse into the interface of coating and substrate [2,10]. The proposed
prone to corrosion damage. The temperature and humidity levels system monitors the mechanical and chemical changes through
inside the museum are controlled and kept at appropriate levels to ␮-strain gauge and ␮LPR sensors. The ␮-strain gauge sensors are
provide suitable environmental conditions to large vehicles. While used to monitor the behaviour of stresses over the coating due to
the harsh atmospheric conditions outside of the museum is consid- change in temperature, whilst the ␮LPR sensors are used beneath
ered as uncontrolled environmental conditions. The ␮LPR sensors the coating to monitor any corrosion reactions that occur due to
embedded on the stationary vehicles in the controlled environment the diffusion of salt particles through micro-cracks. The proposed
have not indicated any significant signs of corrosion, while those structural monitoring system is shown in Fig. 2.
embedded on the stationary vehicles at remote locations in the Consider a coating-substrate system where the applied coating
uncontrolled environment have detected corrosion reactions. has a higher value of coefficient of thermal expansion than the sub-
The residual stresses within the coating-substrate system are strate. In past studies, It was observed that when the temperature
one of the major factors for coating failure in the form of blis- drops, the coating tries to shrink but the substrate, having the lower
tering and micro-crack opening as shown in Fig. 1. The stresses value of coefficient of thermal expansion (CTE) restricts contraction
could be tensile and compressive depending on material properties of the coating. As a result the coating experiences tensile stress
and temperature gradient. According to past experimental study, which opens the crack and allows corrosive particles to diffuse
the coefficient of thermal expansion (CTE) property of coating and through [2], as shown in Fig. 2. As stated, the current sensor suite
substrate is responsible for the occurrence of tensile and compres- contains temperature, strain and ␮LPR sensors, where the tem-
sive stresses, depending on change in temperature [2]. The tensile perature sensor monitors change in temperature, the strain gauge
stress opens the cracks on the coating and allows corrosive par- monitors the development of residual stresses and the ␮LPR sensor
ticles to diffuse into the coating-substrate interface, subsequently detects any corrosion reaction. The complete structural monitor-
resulting in corrosion under the coating as shown in Fig. 1. The ing system consists of a terminal unit with various sensors and
compressive stresses control the number of corrosive particles software at a remote base station.
 J. Latif, Z.A. Khan and K. Stokes / Sensors and Actuators A 301 (2020) 111693 3

Fig. 1. Coating failure and corrosion phenomena.

Fig. 2. Proposed structural monitoring system.

2.1. Remote terminal unit the sensitivity and environmental conditions of the structure. The
external Real Time Clock (RTC) is embedded within the Arduino
Large metal structures are comprised of large and complex Mega 2560 Rev3 and adjusts the time interval for activating the
geometrical configurations, therefore various types of sensors are sensors for measuring and transmitting data to the base station. The
required at critical locations on the structure. Arduino Mega 2560 Global System for Mobile communications (GSM) is used for wire-
Rev3 is used to develop the remote terminal unit (RTU). The Arduino less transmission and allows both stationary and mobile structures
boards are an open source and programmable electronic circuit to be monitored.
board which consists of microcontroller. It can be programmed
using Integrated Development Environment (IDE) software to 2.1.1. Corrosion monitoring technique
design customise embedded systems for various applications. The In this current work, the Linear Polarisation Resistance method
RTU has various channels for corrosion, strain gauge, temperature is used for detecting the corrosion reaction at the interface between
and humidity sensors. The architecture of RTU is described through coating and substrate. Corrosion is an electrochemical process
the flow chart in Fig. 3. which occurs due to the oxidation (anodic) reaction, which involves
The RTU must be synchronised with the current time and date. loss of metal electrons, and the reduction (cathodic) reaction, which
The interval between data sampling is adjusted depending on involves gain of electrons, in the presence of an electrolyte such as
4 J. Latif, Z.A. Khan and K. Stokes / Sensors and Actuators A 301 (2020) 111693

Fig. 3. Flowchart for RTU.

oxygen and water [18]. The oxidation and reduction reactions occur
at random locations on the metal surface in the presence of aque-
ous solution. The oxidation site on the metal results in the loss of
an electron(s) which can be written as:

M(s) −→ M n+ (aq) + ne−

where  M  represents the metal, which is losing ‘n’ number of

electrons. The chemical equation for the cathodic reaction in the
presence of an aqueous solution gaining electrons can be written
as:

nH2 O + ne− −→ nOH − + (n/2)H 2

The ␮LPR sensor is based on three electrode system theory and

includes working, counter/auxiliary and reference electrodes. The
working electrode is in electrical contact with the structure to be
monitored and acts as the anode. The counter electrode completes
the circuit and acts as the cathode. The reference electrode has a
stable known potential. The basic architecture of the ␮LPR is shown
in Fig. 4. The ␮LPR sensor developed by ANALATOM as shown in
Fig. 5 has been utilised for corrosion detection in the current work
and has been previously discussed [13].
Once the value of polarisation resistance is measured through
the ␮LPR sensor, the Stern-Geary equation is applied to calculate
Fig. 4. Three electrode system.
the corrosion rate as follows [15]:

B and cathodic Tafel slopes. The relation for proportionality constant

ICorr = I
Rp ‘B’ is given as:
I  ˇa ˇc
Corr represents the corrosion current, ’Rp is polarisation resis- B= II
tance and ‘B’ is a proportionality constant which depends on anodic 2.303(ˇa + ˇc )
 J. Latif, Z.A. Khan and K. Stokes / Sensors and Actuators A 301 (2020) 111693 5

Fig. 5. Structure of ␮LPR sensor [13].

Fig. 7. Wheatstone bridge with strain gauge sensor.


In the above equation,  ˇa is the anodic Tafel slope and  ˇc is and its resistance changes as it deforms. The change in resistance is
the cathodic Tafel slope. The Tafel slopes can be plotted and mea- measured by connecting the strain gauge into a Wheatstone bridge
sured using ASTM standard G59 [19]. The corrosion rate is then which is used to determine the unknown resistance in the electri-
determined using the following equation: cal circuit. The circuit diagram of the strain gauge connected into
 w  B
 w  the Wheatstone bridge is shown in Fig. 7. The data acquisition unit
Corr Rate = ICorr = III contains a Wheatstone bridge and a strain gauge is connected in a
A*e*F Rp A*e*F
quarter bridge configuration.
The  w represents atomic weight,  F  represents Faraday’s con- Fig. 8 shows the electronic circuit for a ␮-strain gauge sensor
stant,  e is the number of electrons transferred during the corrosion that is connected to the Arduino Mega. The change in strain causes a
process and  A is the area of the corroding electrode. change in resistance, which then results in a change in voltage level.
Fig. 6 shows the electronic circuit for a ␮LPR sensor that is The magnitude of the current is very low, therefore it is amplified
connected to the Arduino Mega. Corrosion occurring on the sen- using the IC-741. The ␮-strain gauge sensor is connected as a resis-
sor causes a change in resistance which then results in a change tance within the Wheatstone bridge and it follows the following
in voltage. As the magnitude of voltage is very low, it requires relationship:
amplification and this is achieved using the IC-741. The gain can
R1 R0
be calculated using the following relationship: = =1 V
R2 Rx
Rf
G =1+ IV The DHT22 sensor as shown in Fig. 9 is used for temperature and
R
humidity measurements. The RTU also monitors battery life using
Where the value of Rf = 100k and R = 10k, respectively. The the electronic circuit shown in Fig. 10.
measured voltage is converted into corrosion current using math- The RTU transmits the data packet to the remote base station
ematical relations programmed in the Arduino Mega. with all required information as shown in Fig. 11. A unique code
acts as an address for the device and validation of the message
2.1.2. Stress monitoring technique from the RTU. It also contains date and time stamps associated with
Stresses in coatings are primarily due to poor coating applica- the sensor measurements. The complete RTU system is shown in
tion and the temperature gradient. The value of the coefficient of Fig. 12.
thermal expansion of coating and substrate defines the magnitude
of stress and strain. The stresses result in deformation of the coat- 2.2. Base station
ing in the form of contraction and widening of micro-cracks. The
strain gauge sensor can be used to monitor the strain produced The software at the base station is configured to the GSM mod-
in the coating due to change in temperature and consists of a foil ule in order to communicate to the RTU’s. The flow chart shown in
pattern which deforms as the coating deforms due to stresses. It Fig. 13 demonstrates how the base station receives a data packet
is embedded on the structure or sample with an adhesive bond from the RTU and processes them for further analysis. The informa-

Fig. 6. Electronic circuit for ␮LPR sensor.

6 J. Latif, Z.A. Khan and K. Stokes / Sensors and Actuators A 301 (2020) 111693

Fig. 8. Electronic circuit for ␮-strain gauge sensor.

Fig. 10. Electronic circuit for battery life.

Fig. 9. Electronic circuit for DHT22 sensor.

ated corrosion environment using salt solution. The sensors were
kept inside the chamber and the response of these sensors and the
tion is extracted from the data packet one by one and operations are wireless communication was monitored at the base station.
performed accordingly. It starts by identifying the structure using a
unique address of RTU. The sensor values associated with time and
date are extracted and stored in the corresponding database of the 3. Experimental setup
structure. The threshold for any critical value can also be defined to
show live critical status on the Graphical User Interface. The com- The proposed sensor-based system was validated through accel-
plete system at the base station which contains software and GSM erated corrosion testing on metal samples coated with primer.
is shown in Fig. 14. Three samples of aluminum 6082 with dimensions 50 mm x 50 mm
In order to verify the wireless communication and response of and thickness 3 mm were prepared. The coefficient of thermal
the sensors, the system was repeatedly tested in salt solution and an expansion of aluminum 6082 is 24 × 10−6 K-1 and Modulus of Elas-
accelerated corrosion environment inside the laboratory as shown ticity is 70 GPa [20]. Sample preparation involved the following
in Figs. 15 and 16. The salt spray chamber provided an acceler- steps:

Fig. 11. Data packet contains all information.

Fig. 12. RTU system with sensors.

 J. Latif, Z.A. Khan and K. Stokes / Sensors and Actuators A 301 (2020) 111693 7

Fig. 15. Inserting a corrosion sensor in salt solution to validate the corrosion detec-
tion.

Fig. 16. Sensors inside the salt spray chamber.

Fig. 13. Flow chart for Base Station.

Table 1
Surface roughness of each sample.

Sample 1Ra(␮m) Sample 2Ra(␮m) Sample 3Ra(␮m)

Average Roughness 1.501 1.197 1.335

• Sample polishing
• Installation of the ␮LPR sensor on the metal substrate prior to
coating
• Application of primer red-oxide coating on samples
• Installation of strain gauge sensor on coated sample

The interfaces of the aluminum 6082 samples were polished

with an emery paper of 80 grit size and are shown in Fig. 17. The
samples were washed with deionised water following polishing.
The roughness of the interfaces after polishing was measured at 8
different points using Optical Inferometery. The average roughness
of each sample is given in the Table 1.
The ␮LPR sensor, with dimensions 40 mm x 20 mm x 0.1 mm,
was then attached to the sample interface ensuring electrical con-
tact to the working electrode of ␮LPR sensor. Adhesive tape was
Fig. 14. Base station system.
used to keep the ␮LPR sensor in place before the coating is applied,
as shown in Fig. 18. The ␮LPR sensor is connected to the RTU which
8 J. Latif, Z.A. Khan and K. Stokes / Sensors and Actuators A 301 (2020) 111693

Fig. 17. Samples after polishing.

Fig. 18. Sample with ␮LPR sensor. Fig. 20. Accessories for installation of ␮-strain gauge sensor.

Fig. 19. Coated sample with ␮LPR sensor.

Fig. 21. Sample with ␮-strain gauge sensor.

converts the electrical signals into digital signals and forwards the After applying the primer, the ␮-strain gauge sensor was
data to the base station. installed on top of the sample as shown in Fig. 20. This requires care-
The red oxide primer, with a coefficient of thermal expansion of ful placing, sticking and soldering of its terminal with wires. The
21 × 10−6 K-1 and a Modulus of Elasticity of 6.14 GPa, was applied coated sample with the attached ␮-strain gauge sensor is shown
on the sample with a conventional spray gun at a temperature of in Fig. 21. A series of experiments were performed in an environ-
300 K. The coated sample was allowed to dry for 24 h before the mental chamber which provided harsh environmental conditions.
next coating was applied. The red oxide primer was applied three The sample was immersed in 10 % salt solution except where the
times and completely covered the sample as shown in Fig. 19. ␮-strain gauge sensor was attached in order to avoid short circuit
 J. Latif, Z.A. Khan and K. Stokes / Sensors and Actuators A 301 (2020) 111693 9

Fig. 22. Sample inside environmental chamber.

Fig. 23. The measurements of ␮-strain gauge sensor during 1st experiment.

problem. The sample along with sensors inside the environmental

chamber is shown in Fig. 22.

4. Results and discussion

Various experiments have been performed to validate the

response of the proposed RTU system with ␮LPR and ␮-strain
gauge sensors. The stress/strain behaviour was observed by provid-
ing small and large temperature gradients. In the first experiment,
the sample was kept in solution to detect the corrosion reaction
under the coating and the development of stresses was observed by
providing a large temperature gradient. The sensor measurements
were recorded continually with a 30 s interval for strain monitor-
ing and 1 min interval for monitoring any corrosion development.
The experiment was started at a temperature of 313.15 ± 2 K and
after 3 h the temperature was decreased to 283.15 ± 2 K. It was
observed that the tensile strain/stress continued to increase as the
temperature was raised from room temperature to 313.15 ± 2 K, as Fig. 24. The measurements of ␮LPR sensor during 1st experiment.
shown in Fig. 23. The initial value of strain recorded was consid-
ered as being an offset that was produced during the application of
the sensor. This offset value was deducted from subsequent strain 1200␮␧ to -200␮␧. The ␮LPR sensor under the coating detects the
measurements in order that the initial measurement started from corrosion reaction and. In the absence of corrosion, it recorded a
zero on the graph. The value of tensile strain started from 0␮␧ and maximum linear polarisation resistance of 5xe6 . The sensor face
reached a maximum of around 1200␮␧. When a large negative tem- connected to the substrate acts as the working electrode. The resis-
perature gradient of T = −30 K was applied, the ␮-strain gauge tance decreases in the presence of corrosion in the substrate. The
sensor experienced compressive strain and a very sharp change in results presented in Fig. 24 show the detection of the corrosion
the behaviour of sensor measurements were observed from around reaction as the sample was dipped into the salt solution. The detec-
10 J. Latif, Z.A. Khan and K. Stokes / Sensors and Actuators A 301 (2020) 111693

Fig. 25. The measurements of ␮-strain gauge sensor during 2nd experiment.

tion of corrosion also confirms the weak coating barrier due to the
presence of micro-cracks that allowed the salt solution to pene-
trate.
In the second experiment, a negative temperature gradient
was applied, although the gradient was low when compared to
the previous experiment. The strain measurements recorded dur-
ing the second experiment are presented in Fig. 25. Initially,
the environmental chamber was set at room temperature for
30 min. and there was no significant change in strain as measured
by the ␮-strain gauge. When a negative temperature gradi-
ent of T = -18k was applied, a compressive strain developed,
which decreased from 0␮␧ to -900␮␧. In addition, a sample was
kept out of the salt solution for the initial 30 min. in order to
observe if any corrosion occurs under the coating in the absence
of the salt solution. No significant corrosion was detected in
these first 30 min., as shown in Fig. 26. Corrosion on the sub-
strate under the coating did start as the sample was dipped
into solution, as shown by the decrease in linear polarisation
resistance.
Fig. 26. The measurements of ␮LPR sensor during 2nd experiment.
In the third experiment, the temperature gradient was applied
in both the positive and negative directions including a small vari-
ation in temperature gradient, as shown in Fig. 27. The initial which is resulted in a slight increase in compressive strain. The tem-
temperature of 292.35 ± 2 K produced a compressive strain and a perature was increased after 2.8 h which resulted in tensile strain
small temperature gradient of T = -3.3 K can also be observed, and the value of strain increased. The corrosion showed interesting

Fig. 27. The measurements of ␮-strain gauge sensor during 3rd experiment.
 J. Latif, Z.A. Khan and K. Stokes / Sensors and Actuators A 301 (2020) 111693 11

5. Conclusion

Experimental investigations have shown that micro coating

defects continuously deteriorate structures, systems and products
in the presence of salt and water. Salinity in atmosphere and rain
are the primary sources of salt and water for structures at remote
locations. The combination of ␮-LPR and ␮-strain gauge sensors
provides an effective solution to continuously monitor the phe-
nomena leading to cororison under the coating. This research has
provided comprehensive experimental investigations and develop-
ment of a structural health monitoring system. This system focuses
on corrosion detection due to diffusion of corrosive particles and
the development of stresses that result in various forms of coat-
ing failure. The threshold levels of failure depend on the properties
of the material of structures, operating conditions and geometrical
configurations. Therefore, experimental methods incorporated all
necessary factors for efficient condition-based maintenance.

Fig. 28. The measurements of ␮LPR sensor during 3rd experiment.

Declaration of Competing Interest

No conflict of interest

Acknowledgments
behaviour which can be linked with the development of stresses
Authors would like to acknowledge financial support provided
within the coating, based on literature [2,21]. During the negative
by Defence Science & Technology Laboratory (DSTL)Ministry of
change in temperature during the first 2.8 h of the experiment, the
Defence (MoD) to conduct this research through grant ID 8983.
corrosion detected was low when compared to that observed after
2.8 h, which had the highest corrosion rate throughout as shown
in Fig. 28. It was concluded that the negative temperature during
References
the cycle before 2.8 h resulted in compressive stresses within the
coating which reduces the diffusion of the salt solution. The posi- [1] 2016; Available from: https://www.nace.org/Newsroom/NACE-News/Study-
tive temperature gradient of T = 8 K, however, resulted in tensile Sets-Course-Toward-Corrosion-Management-Practices-to-Increase-Safety,-
Decrease-$2-5-Trillion-Global-Cost-of-Corrosion/.
strain which expanded the micro-cracks within the coating and
[2] M. Nazir, et al., Modeling the effect of residual and diffusion-induced stresses
allowed greater diffusion of salt solution which resulted in a high on corrosion at the interface of coating and substrate, Corrosion 72 (4) (2015)
corrosion rate. 500–517.
Several layers of coating were applied on substrate which seems [3] M. Nazir, Z.A. Khan, K. Stokes, Optimisation of interface roughness and
coating thickness to maximise coating–substrate adhesion–a failure
to be protecting overall area. The ␮-LPR has detected corrosion prediction and reliability assessment modelling, J. Adhes. Sci. Technol. 29 (14)
under the coating in all three samples due to the penetration of (2015) 1415–1445.
salt solution. It can be considered that even after applying the [4] J. Latif, et al., Condition monitoring and predictive modelling of coating
delamination applied to remote stationary and mobile assets, Struct. Health
coating on large vehicles which are stationed outside in uncon- Monit. (2018).
trolled atmospheric conditions, It is possible that the accumulation [5] M. Nazir, Z.A. Khan, K. Stokes, A holistic mathematical modelling and
of salt particles from atmosphere and rain water can result into simulation for cathodic delamination mechanism–a novel and an efficient
approach, J. Adhes. Sci. Technol. 29 (22) (2015) 2475–2513.
salt solution and penetrate into the interface of coating-substrate [6] J. Latif, et al., Life assessment prognostic modelling for multi-layered coating
system to cause electrochemical reactions and keeps damaging the systems using a multidisciplinary approach, Mater. Sci. Technol. 34 (6) (2018)
structure. 664–678.
[7] M.H. Nazir, Z. Khan, Maximising the interfacial toughness of thin coatings and
Large vehicles exposed to temperature gradient when they
substrate through optimisation of defined parameters, Int. J. Comput.
are often moved from the shed (in Museum) to outside environ- Methods Exp. Meas. 3 (4) (2015) 316–328.
ment for various activities. The temperature gradient responsible [8] M. Nazir, Z. Khan, K. Stokes, Modelling of metal-coating delamination
incorporating variable environmental parameters, J. Adhes. Sci. Technol. 29
in developing residual stresses controls the diffusion of salt parti-
(5) (2015) 392–423.
cles. The current findings has shown the potential of using strain [9] M.H. Nazir, Z.A. Khan, A. Saeed, A novel non-destructive sensing technology
gauge to observe the development of residual stresses subjected for on-site corrosion failure evaluation of coatings, IEEE Access 6 (2018)
to small or large temperature gradients. There are many holis- 1042–1054.
[10] Z.A. Khan, et al., Predictive and prognostic modelling and simulation of
tic and multidisciplinary coating failure and corrosion models coatings subject to corrosion and mechanical failures, Mater. Charact. 6 (3)
can be found in literature, however the real time information (2018) 487–498.
regarding mechanical and electrochemical input parameters to [11] M. Nazir, et al., A predictive model for life assessment of automotive exhaust
mufflers subject to internal corrosion failure due to exhaust gas condensation,
the models is still grey area. The inspection and assessment of Eng. Fail. Anal. 63 (2016) 43–60.
corrosion damage beneath the coating is also very challenging [12] M. Nazir, Z.A. Khan, A review of theoretical analysis techniques for cracking
to schedule the maintenance especially for the structures oper- and corrosive degradation of film-substrate systems, Eng. Fail. Anal. 72 (2017)
80–113.
ating at remote locations. The existence of micro-cracks within [13] D.W. Brown, et al., Linear polarization resistance sensor using the structure as
the coating is not visible during visual inspection. The results a working electrode, in: Proceedings of the Second European Conference of
analysis from proposed SMS has shown that the combination the Prognostics and Health Management Society, Nantes, France, 2014.
[14] S. Harris, M. Mishon, M. Hebbron, Corrosion sensors to reduce aircraft
of ␮-LPR and ␮-strain gauges provides an effective solution for
maintenance, in: Rto avt-144 Workshop on Enhanced Aircraft Platform
real time monitoring of mechanical and electrochemical param- Availability Through Advanced Maintenance Concepts and Technologies,
eters. 2006, Vilnius, Lithuania.
12 J. Latif, Z.A. Khan and K. Stokes / Sensors and Actuators A 301 (2020) 111693

[15] M. Nazir, A. Saeed, Z.A. Khan, Electrochemical corrosion failure analysis of Biographies
large complex engineering structures by using micro-LPR sensors, Sens.
Actuators B Chem. 268 (2018) 232–244.
[16] N. Rossini, et al., Methods of measuring residual stresses in components, Dr. Jawwad Latif received PhD degree in Corrosion health monitoring from
Mater. Des. 35 (2012) 572–588. Bournemouth University. His expertise and research interests are Corrosion moni-
[17] Z.A. Khan, et al., Sensor based corrosion condition monitoring of coating toring system, remote sensing and data compression.
substrate system informed by fracture mechanics, electrochemistry and heat
transfer concepts, in: Department of Defense - Allied Nations Technical Professor Zulfiqar A Khan has developed multidisciplinary research in corrosion,
Corrosion Conference, Birmingham, AL, USA, 2017. corrosion condition monitoring, nano-coating incorporating corrosion and dura-
[18] R.W. Revie, Corrosion and Corrosion Control: An Introduction to Corrosion bility issues, failure analysis, thermodynamics, solid mechanics and mathematical
Science and Engineering, John Wiley & Sons, 2008. modelling.
[19] ASTM G59, Standard Test Method for Conducting Potentiodynamic
Dr Keith Stokes is Visiting Professor within Engineering and Physical Sciences at
Polarization Resistance Measurements, 97, 2014.
the University of Southampton. He has expertise in corrosion, surface engineering,
[20] Aluminium Alloys - Aluminium 6082 Properties, Fabrication and Applications.
microfluidics and high strain body armour.
Available from: https://www.azom.com/article.aspx?ArticleID=2813.
[21] M. Nazir, Z.A. Khan, K. Stokes, Analysing the coupled effects of compressive
and diffusion induced stresses on the nucleation and propagation of circular
coating blisters in the presence of micro-cracks, Eng. Fail. Anal. 70 (2016)
1–15.